278  SMITHSONIAN    MISCELLANEOUS    COLLECTIONS  VOL.    96 
timbers  leaning  slightly  inward  and  the  roof  was  vaulted,  being  formed 
of  whale  ribs,  covered  with  earth.  The  shape  of  the  house  and  the 
form  of  entrance  are  not  mentioned  by  Holmberg,  but  from  Lisi- 
ansky's  description  both  the  inner  room  and  the  sleeping  rooms  seem 
to  have  been  square,  and  Lisiansky,  Langsdorff,  and  Sauer  all  speak 
of  an  entrance  in  the  side,  evidently  through  the  wall.  Benches  or 
platforms  are  not  mentioned,  but  according  to  Holmberg  they  were 
present  in  the  kashims. 
From  de  Laguna's  excavations  (1934,  pp.  157,  158)  we  see  that 
stones  and  whale  bones  were  used  in  the  construction  of  the  oldest 
houses  thus  far  known  in  the  Cook  Inlet  region : 
We  know  nothing  about  the  houses  of  the  First  Period  of  the  Kachemak  Bay 
culture,  and  of  those  of  the  Second  Period  we  know  little,  except  that  large 
stones  and  whale  vertebrae  were  used  in  their  construction.  The  use  of  stone 
in  building  is  anomalous  in  a  region  so  heavily  timbered.  It  could  only  be 
accounted  for  by  the  supposition  that  the  builders  were  people  who  had  come 
from  a  region  where  timber  was  not  plentiful  and  where  stone  was  the  natural 
building  material. 
The  houses  of  the  Third  Period  of  the  Kachemak  Bay  culture  were  semi- 
subterranean — that  is,  they  were  excavated  to  a  depth  of  a  least  2  or  3  feet — 
and  were  entered  by  a  semi-subterranean  entrance  passage,  very  narrow  and  at 
least  12  feet  or  more  long.  The  houses  were  built  of  wood,  with  posts  in  the 
corners  and  at  other  places  along  the  walls.  Some  of  the  houses  had  more  than 
one  room,  and  some  seem  to  have  had  a  fire-place  in  the  main  room. 
There  is  little  information  on  the  houses  of  the  southernmost  group 
of  Eskimos,  those  at  Prince  William  Sound.  Petrofif  (1884,  p.  28) 
remarks  only  that  they  were  underground,  a  statement  which  seems 
borne  out  by  Steller's  brief  description  of  a  house  (or  cache?)  that 
he  saw  on  Kayak  Island  in  1741. 
I  pushed  the  grass  aside  at  once  and  found  underneath  a  cover  consisting  of 
rocks ;  and  when  tliis  was  also  removed  we  came  to  some  tree  bark,  which  was 
laid  on  poles  in  an  oblong  rectangle  three  fathoms  in  length  and  two  in  width. 
All  this  covered  a  cellar  two  fathoms  deep.    [Colder,  1922-25,  vol.  2,  p.  48.] 
The  houses  of  the  Northwest  Coast  Indians  present  a  special 
problem,  the  solution  of  which  will  require  a  more  comprehensive 
and  searching  analysis  than  has  as  yet  been  made.  The  presence 
of  an  inner  pit  would  seem  to  indicate  a  basic  relationship  with  the 
semisubterranean  earth  covered  houses  to  the  north  and  south,  as 
Waterman  (1921,  pp.  23-30)  has  pointed  out.  But  in  many  respects 
the  Northwest  Coast  houses  have  developed  so  far  beyond  these 
simpler  and  undoubtedly  more  primitive  forms  that  they  constitute 
a  definite  break  in  the  line  of  such  houses.    In  spite  of  this,  it  is  here 
